OPINION
This consolidated action returns to court,2 following remand3 and redetermi-nation4 of the final results of the anti-dumping duty investigation of multilayered wood flooring from the People’s Republic of China (“PRC” or “China”).5 Plaintiffs, cooperative non-investigated respondents who have established their entitlement to a separate antidumping duty rate, challenge the remand redetermination of that rate.6 Plaintiffs argue that the Department of Commerce’s (“the Department” or “Commerce”) redetermination is flawed because the Department’s legal interpretations are not in accordance with the law and the Department’s factual conclusions are not supported by a reasonable reading of the evidence.7
Plaintiffs are, in part, correct. Commerce has not articulated a rational connection between the record evidence and the rate applied to the separate rate companies, nor has Commerce explained how its determination bears a relationship to Plaintiffs’ economic reality. Accordingly, the court remands to Commerce for further consideration in accordance with this opinion.
BACKGROUND
I. Baroque III
This dispute originates in a petition by the Coalition for American Hardwood Par*1337ity (“CAHP”) alleging that imports of mul-tilayered wood flooring from the PRC were being dumped in the United States. In response, Commerce initiated an anti-dumping duty investigation for the period of April 1, 2010 through September 30, 2010. Multilayered, Wood Flooring from the People’s Republic of China, 75 Fed. Reg. 70,714 (Dep’t Commerce Nov. 18, 2010) (initiation of antidumping duty investigation) (“Initiation Notice ”). Commerce indicated that it would select mandatory respondents based on quantity and value (“Q & V”) questionnaires. Id. at 70,717. Commerce requested Q & V data from 190 companies and received timely responses from 80. Multilayered Wood Flooring from the People’s Republic of China, 76 Fed.Reg. 30,656, 30,657 (Dep’t Commerce May 26, 2011) (preliminary determination of sales at less than fair value) (“Preliminary Determination ”). From these, Commerce selected three mandatory respondents, the largest cooperating exporters (by volume) of wood flooring, for the investigation: Zhejiang Yuhua Timber Co., Ltd. (“Yuhua”), Zhejiang Layo Wood Industry Co., Ltd. (“Layo”), and the Saml-ing Group8 (“Samling”). Id. at 30,658; see also 19 U.S.C. § 1677f-l(c)(2)(B).9 Those companies that failed to respond to Commerce’s Q & V questionnaire were treated as part of the PRC-wide entity. Preliminary Determination at 30,661.10
In addition, because this was a non-market economy (“NME”) investigation,11 Commerce invited those exporters and producers seeking a separate rate to submit a separate-rate status application.12 Commerce received timely-flled separate-rate applications from 74 companies, all of which demonstrated eligibility for separate rate status. Final Determination at 64,-321.13
*1338In its Final Determination, Commerce found that multilayered wood flooring was being dumped in the United States. Id. at 64,323-24. Commerce found a de minimis dumping margin for Yuhua and assigned margins of 3.98 percent and 2.63 percent to Layo and Samling, respectively. Id. Commerce assigned the AFA rate of 58.84 percent (the highest calculated transaction-specific rate among mandatory respondents) to the PRC-wide entity. Id. at 64,322. Commerce then assigned the separate rate respondents a rate of 3.31 percent. Id. This rate was the simple average of Layo and Samling’s margins. I & D Memo, cmt. 11 at 51.14
Plaintiffs sought judicial review of the Final Determination pursuant to 19 U-S.C. §§ 1516a(a)(2)(A)(i)(II) and 1516a(a)(2)(B)(i), and Commerce requested a voluntary remand. The court affirmed in part and remanded in part. The court affirmed Commerce’s rejection of Respondents’ late filed surrogate financial statements. The court remanded to Commerce for reconsideration the surrogate value (“SV”) determinations for Layo’s plywood input and Samling’s HDF input; remanded for reconsideration Commerce’s targeted dumping determination, in light of any changes to the surrogate value determinations and current standards; and remanded for further explanation or reconsideration the surrogate value determination for Layo’s core veneer, Layo’s HDF input, and Layo’s brokerage and handling (“B & H”) fees to account for the cost of a letter of credit. Baroque III, — C.I.T. at-, 925 F.Supp.2d at 1337; see also Remand Results at 1-2.
II. Commerce’s Redetermination Pursuant to Remand
In its Redetermination, Commerce revised its findings as required by Baroque III. Commerce (1) valued Layo’s plywood input with an SV reflecting plywood thicknesses of 6.35 mm and 12.7 mm; (2) valued Samling’s high-density fiberboard (“HDF”) with Philippine Harmonized Tariff Schedule (“HTS”) category 4411.11; (3) valued Layo’s core veneer input with 2009 data reported by the Global Trade Atlas for Philippine HTS category 4408.9090.06; (4) provided further explanation for Commerce’s determination “to continue converting SV for [Layo’s] HDF using the average density of HDF used by [Layo]”; (5) adjusted Layo’s “B & H SV to remove letter of credit costs not incurred by [Layo]”; and, (6) calculated Layo’s and Samling’s dumping margins “using an average-to-average comparison method, rather than the average-to-transaction comparison method.” Remand Results at 2.
As a result of these changes, not only Yuhua, but also Layo and Samling received dumping margins of zero. Id. at 26.15 The changes to Layo and Samling’s *1339SVs resulted in a new calculated highest transaction-specific rate of 25.62 percent. Commerce selected this rate as the revised AFA rate for the PRC-wide entity. Id. at 27. Because all the mandatory rates were zero, Commerce chose to recalculate the separate rate under 19 U.S.C. § 1673d(c)(5)(B)’s “any reasonable method provision,” taking a simple average of the three mandatory rates of zero and the AFA rate. This resulted in a separate rate of 6.41 percent, id., thereby increasing the separate respondents’ rate while each of the components of that rate decreased.
STANDARD OF REVIEW
The court will uphold Commerce’s determinations unless they are “unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(l)(B)(i).
Substantial evidence is “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Universal Camera Corp. v. N.L.R.B., 340 U.S. 474, 477, 71 S.Ct. 456, 95 L.Ed. 456 (1951) (quoting Consol. Edison Co. of New York v. N.L.R.B., 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938)). It must be “more than a scintilla, and must do more than create a suspicion of the existence of the fact to be established.” N.L.R.B. v. Columbian Enameling & Stamping Co., 306 U.S. 292, 300, 59 S.Ct. 501, 83 L.Ed. 660 (1939). In making its judgment, the court “looks to the record as a whole, including any evidence that fairly detracts from the substantiality of the evidence,” Gallant Ocean (Thailand) Co., Ltd. v. U.S., 602 F.3d 1319, 1323 (Fed.Cir.2010) (internal quotation marks and citation omitted),16 and determines “whether the evidence and reasonable inferences from the record support [the agency’s] finding.” Daewoo Elees. Co. v. Int’l Union of Elec., Elec., Technical, Salaried & Mach. Workers, AFL-CIO, 6 F.3d 1511, 1520 (Fed.Cir.1993) (internal quotation marks and citation omitted).17 Commerce must provide a “rational connection between the facts found and the choice *1340made.” Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 168, 83 S.Ct. 239, 9 L.Ed.2d 207 (1962). It must “examine the record and articulate a satisfactory explanation for its action.” Yangzhou Bestpak Gifts & Crafts Co., Ltd. v. United States, 716 F.3d 1370, 1378 (Fed.Cir.2013).
In essence, the substantial evidence standard asks whether Commerce’s determination was reasonable. Nippon Steel, 458 F.3d at 1351 (quoting SSIH Equipment SA v. United States ITC, 718 F.2d 365, 381 (Fed.Cir.1983) (Nies, J. additional comments)).18
DISCUSSION
I. Commerce’s Methodology
A. The statutory provision allows Commerce to use “any reasonable method.”
Otherwise lacking statutory guidance,19 Commerce follows 19 U.S.C. § 1673d(c)(5) (method for determining the estimated all-others rate) when calculating the dumping margin for separate rate respondents. Remand Results at 45.
Section 1673d(c)(5)(A) provides the general rule,20 but when all of the weighted average dumping margins for individually investigated exporters and producers are zero, de minimis, or based entirely on facts available, an exception, Section 1673d(c)(5)(B), applies and Commerce may use “any reasonable method” to establish the separate rate.21
Here, because on remand the mandatory respondents all had weight-averaged dumping margins of zero, Commerce calculated the separate rate margin under the Section 1673d(c)(5)(B) “any reasonable method” provision. Remand Results at 45. Commerce took a simple average of the three mandatory respondent zero rates and the PRC-wide AFA rate. Id.
*1341 B. It is not per se unreasonable for Commerce to use a simple average of zero and AFA rates to calculate the separate rate.
Section 1673d(c)(5) does not say whether a simple average of three zero percent mandatory respondent rates and the PRC-wide AFA rate is reasonable. Because the statute does not “directly address the precise question at issue,” the court is left to decide whether Commerce’s interpretation is “a reasonable construction of the statute.” Bestpak, 716 F.3d at 1377.22
Section 1673d(c)(5)(B)’s breadth and flexibility allow for a contextual application of the statute.23 It follows that there is “no legal error” inherent in using a simple average rather than a weighted average. Bestpak, 716 F.3d at 1378. And, as both “[Section] 1673d(c)(5)(B) and the SAA explicitly allow Commerce to factor both de minimis and AFA rates [of individually investigated exporters and producers] into the calculation methodology.” Id. Accordingly, as a method “derived from the relevant statutory language,” id. at 1378, it is not per se unreasonable for Commerce to use a simple average of de minimis and AFA rates to calculate the separate rate antidumping duty margin.24
II. Commerce’s Method is Not Supported by Substantial Evidence
A. Commerce’s chosen method must be reasonable as applied in order to be supported by substantial evidence.
While Commerce’s chosen method may not be per se unreasonable, it must still be reasonable as applied.25 In order for an antidumping duty determination to be reasonable as applied, Commerce must articulate a “rational connection between the facts found and the choice made.” Burlington Truck Lines, 371 U.S. at 168, 83 S.Ct. 239. Commerce must “examine the record and articulate a satisfactory explanation for its action.” Bestpak, 716 F.3d at 1378.26 At the very *1342least, it must “cogently explain why it has exercised its discretion in a given manner.” Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 48, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983).27 A determination cannot be considered reasonable if the agency has “entirely failed to consider an important aspect of the problem” before it. Id. at 43, 103 S.Ct. 2856.
When the problem is dumping, any method Commerce employs must be “based on the best available information and establish ] antidumping margins as accurately as possible.” Shakeproof Assembly Components, Div. of Ill. Tool Works, Inc. v. United States, 268 F.3d 1376, 1382 (Fed.Cir.2001).28 While 19 U.S.C. § 1673d(c)(5)(B) allows Commerce to use “any reasonable method,” it must be in service of calculating a margin “reasonably reflective of potential dumping margins for non-investigated exporters or producers.” 1994 U.S.C.C.A.N. 4040, 4201.29
Because judicial review of an administrative decision must be made on the grounds relied on by the agency,30 if Com*1343merce has not articulated its reasoning sufficiently, the court will require “such additional explanation of the reasons for the agency decision as may prove necessary.” Camp v. Pitts, 411 U.S. 138, 142-43, 93 S.Ct. 1241, 36 L.Ed.2d 106 (1973).
B. Commerce failed to articulate a rational connection between facts found and choices made.
In its Redetermination, Commerce did not consider whether use of an AFA rate, let alone use of the selected transaction-specific margin, was merited in its separate rates calculation. Nor did Commerce consider its responsibility to determine a separate rate that bears some relationship to respondents’ actual rates. Rather, Commerce explains that its use of the AFA rate in the separate rate calculation is reasonable because Commerce needed to account for the non-cooperating, PRC-wide companies in the investigation. Remand Results at 46. Because some companies refused to respond to Commerce’s requests for Q & V data, Commerce correctly notes that it lacks a complete data set. Commerce suggests that because any of the non-cooperating companies could or “may have been selected” as a mandatory respondent, Commerce must account for them in some way in the separate rate calculation. Commerce suggests that it cannot be sure that the mandatory respondents are reflective of the separate rate respondents. Id. 31
While Commerce may draw reasonable inferences from the failure of uncooperative respondents to provide evidence of the size, quantity, and value of their sales, doing so does not provide a rationale for the redetermination made here. The mere presence of non-cooperating parties “fails to justify [Commerce’s] choice of dumping margin for the cooperative uninvestigated respondents.” Amanda Foods (Vietnam) Ltd. v. United States, 33 C.I.T. 1407, 647 F.Supp.2d 1368, 1381 (2009) (“Amanda Foods I”).
Application of the AFA rate to non-cooperating parties is a rebuttable presumption. See Rhone Poulenc, 899 F.2d at 1190-91. A rebuttable presumption is not evidence. New York Life Ins. Co. v. Gamer, 303 U.S. 161, 170, 58 S.Ct. 500, 82 L.Ed. 726 (1938).32 Even if it were, the fact that the AFA rate applies to other companies is not evidence of dumping on the part of the separate rate companies. Amanda Foods I, 647 F.Supp.2d at 1381. Commerce cannot use the AFA rate in calculating the separate rate for cooperating parties without explanation. See Changzhou Wujin Fine Chem. Factory, 701 F.3d at 1379.
Moreover, Commerce failed to make any connection between the transaction-specific margin of 25.62 percent and separate rate respondents’ pricing practices. Com*1344merce did not provide a rationale for how its use of this margin results in a reasonably accurate separate rate. While Commerce’s concern about incomplete Q & V data provides an explanation for its decision to use a method other than the expected average of individually investigated rates, that rationale has no relationship to the use of the 25.62 percent transaction-specific margin. Why, for example, would it not have been appropriate to include a different or multiple transaction-specific margins in order to get a more accurate rate? Specifically, why this margin? How has Commerce done other than “cherry-picked [a] single data point” and gratuitously added it to the separate rate calculation?33 Why, on the factual record, is this a reasonable way for Commerce to have exercised its discretion? The Rede-termination contains no consideration of this aspect of the problem.
It is, of course, correct that, to calculate the separate rate in the Redetermination, Commerce has moved from (a modified application of) the general rule of 19 U.S.C. § 1673d(c)(5)(A) to the exception in 19 U.S.C. § 1673d(c)(5)(B), reflecting changes in the mandatory rates. But Commerce has failed to consider its responsibility to determine rates that bear some relationship to respondents’ actual rates, to their economic reality, rendering its chosen method unreasonable. Whether under the general rule or the exception, the mandatory respondents are meant to be representative of the industry, and therefore of the separate rate respondents. See 19 U.S.C. § 1677f-l(c)(2).34 Even under the exception, which allows for “any reasonable method,” the expected method is an average of the “estimated weighted average dumping margins determined for the exporters and producers individually investigated.” 19 U.S.C. § 1673d(c)(5)(B).35 Commerce has exercised its discretion to not use the expected method in favor of a method that takes into account the absence of data from the PRC-wide entity. While the use of the AFA rate in the *1345calculation of the separate rate may be reasonable in some circumstances (so long as supported by substantial evidence), here the seemingly gratuitous inclusion of this transaction-specific rate in the separate rate calculation, to increase the resultant rate, is incongruous. Upon remand, all relevant rates — mandatory, transaction-specific and AFA — decreased, suggesting a decreased likelihood of dumping.36 But Commerce made the choice to use a method that increased the separate rate both from the zero that would have resulted from the expected method and from the 3.31 percent in Commerce’s original determination. Commerce did not explain why it made this choice or how the result was in any way reasonably reflective of Plaintiffs’ economic reality.37
While it is true that under substantial evidence the court “do[es] not make the determination,” it “merely vet[s] the determination,” Nippon Steel, 458 F.3d at 1352, “that the scope of such review is narrowly circumscribed is beside the point,” Chenery, 318 U.S. at 94, 63 S.Ct. 454, where, as here, Commerce’s redetermination fails to articulate the required rational connection between the facts found and the rate chosen. It therefore fails substantial evidence review.
*1346 CONCLUSION
It is lawful for Commerce to draw reasonable inferences from uncooperative companies’ failure to submit evidence of the size, quantity, and value of their sales, and to use a method reasonably derived from the relevant statutory language. But substantial evidence asks a more specific question, and requires a more specific explanation from Commerce.38 At issue is whether Commerce’s determination was based on a reasonable reading of the record in context. Without further explanation, the court cannot consider it so.39
Accordingly, this matter is affirmed in part and remanded in part to Commerce for further consideration in accordance with this opinion. Commerce shall have until May 8, 2014 to complete and file its remand redetermination. Plaintiffs shall have until May 22, 2014 to file comments. Defendant and Defendant-Intervenors shall have until June 6, 3014 to file any reply.
IT IS SO ORDERED.